Thanks, the aliasing is definitely not wrapping the spectrum correctly.
What should be happening is that the 0th row is repeated after the (N-1)st
row so that the contouring is completed, and it seems fairly obvious from
those pictures that it is not doing that, so I'll have to change that.
(This kind of boundary fix happens internally in between blocks inside one
tiling, but it needs to be extended to the proper boundary of the tile as
well.) The same fix will need to be done to the peak finding because
there is the same issue there. That is why you are getting two peaks, it
is hitting the boundary in both cases.
There is a slightly subtle point here. Sometimes part of the spectrum is
thrown out, for example in Azara process using the "range" command. In
this case the contours should not be wrapped. The data model caters for
this issue but it further complicates the code. One of the problems with
importing Azara, and possibly other, data is that there is no record of
whether the "range" command has been done, so Analysis currently assumes
it has not, which is obviously wrong in many cases, and that will cause
noticeable artefacts when the above bug is fixed. When Azara process gets
the data model wrapped around it this will no longer be an issue for that
specifically. But because a lot of data imported into Analysis is not
from Azara, I think we will probably put a "wrap spectra" flag into the
code or allow the user to specify the original spectral width, etc.
Wayne
On Fri, 9 Jun 2006, Ben Goult wrote:
> Hi
>
> I have just unaliased a folded dataset but it doesnt look quite right
> using Analysis.
>
> When I look at the "unfolded" spectrum there is a gap between the edge
> of the actual original spectrum and the start of the next tile of the
> spectrum. I have enclosed a snapshot to illustrate this as its hard to
> describe.
>
> When I peak pick the spectrum I get two peaks picked for each peak that is
> located on the edge of the spectrum, one for each half of the peak.
>
> Does anyone know what is wrong with it or how I can correct it?
>
> Cheers
>
> Ben
>
|